2.6. Inequality Data Analysis¶
%matplotlib inline
import numpy as np
import matplotlib.pyplot as plt
import seaborn as sns; sns.set()
import pandas as pd
import plotly.plotly as py
import cufflinks as cf
import plotly.graph_objs as go
from sklearn.cluster import KMeans
from sklearn.preprocessing import LabelEncoder
from sklearn.preprocessing import MinMaxScaler
from scipy.stats import pearsonr, spearmanr
%matplotlib inline
cf.set_config_file(theme='space',offline=True)
pd.set_option('display.max_colwidth', -1)
2.6.1. Data Importing¶
df = pd.read_csv("../data/production/subject/Inequality.csv").set_index(["Country Code","Year"])
dd = pd.read_csv("../data/production/data_dictionary.csv").set_index("Code").loc[df.columns]
tourism_columns = ['ST.INT.ARVL', 'ST.INT.XPND.MP.ZS', 'ST.INT.XPND.CD', 'ST.INT.DPRT',
'ST.INT.RCPT.XP.ZS', 'ST.INT.RCPT.CD', 'Tourist Defecit', 'Tourism Net',
'Tourist Avg Net', 'Population Estimate', 'ST.INT.ARVL.PER.CAPITA',
'ST.INT.DPRT.PER.CAPITA']
df.drop(["population","gdp_ppp_pc_usd2011","id","quality_score"],axis="columns",inplace=True)
/opt/conda/lib/python3.7/site-packages/ipykernel_launcher.py:2: FutureWarning:
Passing list-likes to .loc or [] with any missing label will raise
KeyError in the future, you can use .reindex() as an alternative.
See the documentation here:
https://pandas.pydata.org/pandas-docs/stable/indexing.html#deprecate-loc-reindex-listlike
dd
| Indicator Name | |
|---|---|
| Code | |
| ST.INT.ARVL | International tourism, number of arrivals |
| ST.INT.XPND.MP.ZS | International tourism, expenditures (% of total imports) |
| ST.INT.XPND.CD | International tourism, expenditures (current US$) |
| ST.INT.DPRT | International tourism, number of departures |
| ST.INT.RCPT.XP.ZS | International tourism, receipts (% of total exports) |
| ST.INT.RCPT.CD | International tourism, receipts (current US$) |
| Tourist Defecit | The difference in outbound-inbound tourists for a country |
| Tourism Net | The difference in tourism recepts-expenditures |
| Tourist Avg Net | The average net income per tourist |
| Population Estimate | The UNPD estimated population for the country |
| ST.INT.ARVL.PER.CAPITA | Inbound tourists per resident |
| ST.INT.DPRT.PER.CAPITA | Outbound tourists per resident |
| SI.POV.GINI | GINI index (World Bank estimate) |
| SI.POV.GINI | GINI index (World Bank estimate) |
| SI.DST.10TH.10 | Income share held by highest 10% |
| SI.DST.FRST.10 | Income share held by lowest 10% |
| id | Identifier |
| gini_reported | Gini coefficient as reported by the source (in most cases based on microdata, in some older observations estimates derive from grouped data) |
| q1 | Quintile group shares of resource |
| q2 | Quintile group shares of resource |
| q3 | Quintile group shares of resource |
| q4 | Quintile group shares of resource |
| q5 | Quintile group shares of resource |
| d1 | Decile group shares of resource |
| d2 | Decile group shares of resource |
| d3 | Decile group shares of resource |
| d4 | Decile group shares of resource |
| d5 | Decile group shares of resource |
| d6 | Decile group shares of resource |
| d7 | Decile group shares of resource |
| d8 | Decile group shares of resource |
| d9 | Decile group shares of resource |
| d10 | Decile group shares of resource |
| mean | Survey mean given with the same underlying definitions as the Gini coefficient and the share data |
| median | Survey median given with the same underlying definitions as the Gini coefficient and the share data |
| exchangerate | Conversion rate from local currency units (LCU) to United States Dollars (USD) |
| gdp_ppp_pc_usd2011 | Gross Domestic Product (GDP) is converted to United States Dollars (USD) using purchasing power parity rates and divided by total population. Data are in constant 2011 United States Dollar (USD) |
| population | NaN |
| quality_score | NaN |
2.6.2. Correlations¶
corr = df.corr().drop(tourism_columns,axis="columns").loc[tourism_columns]
corr.iplot(kind='heatmap',colorscale='-rdbu',
filename='economic-heatmap',
title="Correlations between Tourism indicators and economic",
zerolinecolor="white",
dimensions=(640,500),margin=(150,150,150,50))
2.6.3. Variable Distributions¶
def draw_histograms(dataframe):
ax = dataframe.iplot(kind='histogram', subplots=True, shape=(9,4))
return ax
draw_histograms(df)
2.6.4. Analysis¶
2.6.4.1. Inequality Correlation with Tourism Variables¶
2.6.4.1.1. A Bigger market share for tourism results in higher equality¶
\(H_0: \rho = 0\) >There is not a significant linear relationship between x and y
\(H_a: \rho \neq 0\) >There is a significant linear relationship between x and y
\(\alpha = 0.01\)
x = df_norm['ST.INT.RCPT.XP.ZS']
y = df_norm['SI.POV.GINI']
c = spearmanr(x,y)
print("The two variables have a spearman correlation of {} with a pvalue of {}.".format(c.correlation,c.pvalue))
The two variables have a spearman correlation of 0.007586460697844305 with a pvalue of 0.6066161826004317.
\(\rho \equiv 0\) at \(\alpha = .01\)
We fail to reject the null hypothesis
plt.scatter(x,y)
plt.xlabel("International tourism, receipts (% of total exports)")
plt.ylabel("GINI index");
2.6.4.1.2. A Bigger market share for tourism results in higher equality (round 2)¶
\(H_0: \rho = 0\) >There is not a significant linear relationship between x and y
\(H_a: \rho \neq 0\) >There is a significant linear relationship between x and y
\(\alpha = 0.01\)
x = df_norm['ST.INT.RCPT.XP.ZS']
y = df_norm['SI.DST.FRST.10']
c = spearmanr(x,y)
print("The two variables have a spearman correlation of {} with a pvalue of {}.".format(c.correlation,c.pvalue))
The two variables have a spearman correlation of 0.05641561711062179 with a pvalue of 0.00012708108793161872.
\(\rho = 0\) at \(\alpha = .01\)
The null hypothesis is rejected, and the variables are slightly correlated.
plt.scatter(x,y)
plt.xlabel("International tourism, receipts (% of total exports)")
plt.ylabel("Income share held by lowest 10%");
2.6.4.1.3. A Bigger market share for tourism results in lower equality¶
\(H_0: \rho = 0\) >There is not a significant linear relationship between x and y
\(H_a: \rho \neq 0\) >There is a significant linear relationship between x and y
\(\alpha = 0.01\)
x = df_norm['ST.INT.RCPT.XP.ZS']
y = df_norm['SI.DST.10TH.10']
c = spearmanr(x,y)
print("The two variables have a spearman correlation of {} with a pvalue of {}.".format(c.correlation,c.pvalue))
The two variables have a spearman correlation of 0.04620014371945418 with a pvalue of 0.0017049151198087406.
\(\rho = 0\) at \(\alpha = .01\)
The null hypothesis is rejected, and the variables are slightly correlated.
plt.scatter(x,y)
plt.xlabel("International tourism, receipts (% of total exports)")
plt.ylabel("Income share held by richest 10%");
